Section 230: Redefining The Modern Public Square

November 17th, 2025

---

---

In the fledgling digital age, the internet was the new Final Frontier - few laws, government departments, or even companies were prepared to address specific new issues that arose with it. The Communication Decency Act (CDA) was passed into law as Section V of the Telecommunications Act, signed by President Bill Clinton on February 8th, 1996 (Telecommunications Act of 1996, Pub. L. No. 104-104, § 502, 110 Stat. 56, 133–35). This set of regulations was meant to be an attempt to “clean up” all forms of digital media, defining and addressing various issues within online and other forms of communication. The CDA specifically addressed the transmission of “obscene, lewd, lascivious, filthy…” (47 U.S.C. § 230(c)(2)) and several other terms for offensive media as a whole. Section 230, often known as the Good Samaritan section, is the only remaining legislation from the original act, which has since been updated, amended, and had all but this section struck down through cases such as Reno v. ACLU and Ashcroft v. Free Speech Coalition. 

Reno v. ACLU was the primary agent in the CDA’s downfall – the court ruled against a large part of the new legislation, holding that it was “facially overbroad in violation of the First Amendment…” and “were content-based blanket restrictions on speech… could not be properly analyzed on First Amendment challenge as a form of time, place, and manner regulation“ (Reno v. ACLU, 521 U.S. 844 (1997)). Originally crafted to protect the general public – and specifically, minors – from obscene or otherwise corrupting material, this remaining section has been abused on more than one occasion as protection for modern internet providers and social media giants’ selective censorship (Telecommunications Act of 1996, 1996; 521 U.S. 844, 117 S. Ct. 2329, 138 L. Ed. 2d 874; Ashcroft v. Free Speech Coal., 535 U.S. 234 (2002)).

In its current form, the first subsection of Section 230(c) now classifies all providers of apps and websites as non-publishers, detailing that they cannot be held liable for anything posted on their platforms (47 U.S.C. § 230(c)(1)). This is commonly referred to as the public square - a place where any content from users is allowed to be posted at will. However, it also includes protection for removal or censorship of content mentioned above. The second subsection of 230(c) states that “any action voluntarily taken in good faith” to restrict access to content that the publisher considers “obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected” (47 U.S.C. § 230(c)(2)) is fully protected from any liability in lawsuits that might arise pertaining to the First Amendment. There are several key elements to this section that are a topic of frequent discussion – the “constitutionally protected” phrase, the classification of “otherwise objectionable”, and the definition of a publisher itself. To analyze this section properly, however, one must first gain an understanding of the context and intent surrounding its creation. 

The original CDA was created in 1996, when the internet had approximately 40 million users and was, as expected, somewhat of a cesspool of pornography, dangerous information, and rampant defamation (NTIA, Falling Through the Net: A Survey of the “Have Nots” in Rural and Urban America, U.S. Dep’t of Commerce (July 1995). A senate report in 1996 detailed the “increasing number… inappropriate uses of telecommunications technologies to transmit pornography, engage children in inappropriate adult contact, terrorize… users through ‘‘electronic stalking,’’ and seize personal information” (S. Rep. No. 104-23, §B(8), 1995). A major element that contributed to the state of the web at that time hinged on two cases that essentially terrified most internet platforms into inaction.

The first of these was Cubby v. CompuServe, a landmark online service founded in 1969 that provided email, forums, file sharing, and several other major features most often associated with Google today. In this case, Cubby, Inc sued CompuServe for its “alleged libel, business disparagement, and unfair competition” (Cubby, Inc. v. CompuServe, Inc., 776 F. Supp. 135 (S.D.N.Y. 1991)) after a defamatory newsletter from an online magazine titled “Rumorville USA” accused Cubby, Inc. of being a scam run by criminal individuals. After review, the court granted summary judgment in favor of CompuServe, holding that an “utterance of a disparaging statement must be intentional” for it to be “held liable on plaintiffs' unfair competition claim if it did not know or have reason to know of [these]... statements” (Cubby, Inc. v. CompuServe, Inc., 776 F. Supp. 135 (S.D.N.Y. 1991)). This demonstrated quite quickly to other companies carrying similar content on the internet that protection hinged on non-interference – CompuServe was not the author, nor the publisher of the content found on its site, and therefore could not be held liable for the content therein. However, the definition of publisher was left somewhat up for debate until Stratton Oakmont, Inc. v. Prodigy Services Co. happened just four years later in 1995.

Prodigy Services Corporation was a groundbreaking internet service founded in 1988 by several major corporations (IBM, Sears, CBS, and Roebuck & Co.) that essentially created its own internet before a truly connected network appeared in the United States. They provided news, weather, forums, and ecommerce within their own proprietary environment, accessed via exclusive dial-up routers. However, what set them apart from similar services is that they were the first major network to promote content moderation, advertising themselves as a family-friendly network with strict auto-moderation features and manual content review. Stratton Oakmont, Inc. filed for defamation when a forum post accused them of, once again, being criminals and frauds.  Prodigy removed the posts per Stratton’s request, but would not pay for damages or apologize – hence the lawsuit. However, the court’s ruling took a very different approach due primarily to a concept known as “editorial control”.

The decision stated that “PRODIGY's conscious choice to gain the benefits of editorial control” decidedly opened it to “greater liability than CompuServe and other computer networks that [made] no such choice” (Stratton Oakmont, Inc. v. Prodigy Servs. Co., 1995 WL 323710 (N.Y. Sup. Ct. May 24, 1995)).  Due to the fact that Prodigy chose which content was allowable and which wasn’t, and that they advertised their service as family friendly (curating content), the court officially determined that the organization would be held to the same standard as a publisher – thus, they were liable for any and all content created or posted on their site. This essentially served as an immediate notice to all other corporations with similar products that a fully “hands off” approach was the only definitive way to avoid legal liability, which led to a host of ethical questions about what sort of illegal activity should be allowed on the internet, and what role providers would take in the execution of crimes and other morally objectionable use of the web. As this question rose to the forefront of lawmakers’ minds, the CDA – and Section 230 – were conceived.

Instead of seeking to create government regulation and to codify specific requirements for content, likely triggering a variety of lawsuits and challenges to First Amendment rights, lawmakers simply enacted a good-faith policy (Brannon, V.C., 2024). To establish a distinct classification for protected parties, the act stated that “no provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider” (47 U.S.C. § 230(c)(1)). In practical terms, this established that service providers officially could not be held liable for any content posted by a user on their platform. However, this only codified the outcome of the first of the two key cases – to address this, a second subsection specifically addressing civil liability was added. The legislators determined that “any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected” (47 U.S.C. § 230(c)(2)). This served to address the second court decision, clarifying that not only could platforms not be held liable for inaction, but they were protected almost universally when taking action as well.

It is at this point where several questions arise. While the act continues on to clarify that this new regulation would not expand or limit other laws such as privacy, intellectual property, sex trafficking, and several others, it does not expand upon the meaning of “otherwise objectionable”, nor does it clarify how “constitutionally protected” material can be deemed objectionable within a public square platform and removed without potential ramifications. This set of laws was written in a society markedly different from today’s. The expectation was that companies and platforms would use these rules sparingly, only removing content that could be deemed universally harmful. However, the loophole of “otherwise objectionable” has recently been exploited on a significant scale, seen particularly during COVID and recent elections. One of the more notable cases where § 230 truly came under legal scrutiny for the first time occurred during a 2021 lawsuit, Gonzalez v. Google LLC. Several family members of victims in a series of ISIS terrorist attacks brought suit against Google, Twitter, and Facebook. The plaintiffs stated that these platforms had abetted and assisted these attacks by allowing members of ISIS to post recruitment, planning, and radicalization messages without consequence to the platform, and that paid advertisements close to the content shared revenue with the terrorist group (Gonzalez v. Google LLC, 2 F.4th 871 (9th Cir. 2021)).

The case was dismissed by a lower court and affirmed by the 9th Circuit Court of Appeals before being taken to the Supreme Court, where it was eventually vacated and remanded due to the decision made in another simultaneous case on nearly the same issue, Twitter v. Taamneh (Twitter, Inc. v. Taamneh, 598 U.S. 471, 143 S. Ct. 1206, 215 L. Ed. 2d 444 2023)). The court ruled unanimously that each and every one of the defendants were not liable for the terrorist attacks in any way, shape, or form, reversing a lower court decision. The court held that “plaintiffs failed to sufficiently allege that operators of social media platforms consciously and culpably participated in [the] Istanbul attack” (Twitter, Inc. v. Taamneh, 598 U.S. 471 (2023)). This established direct precedent and furthered the liability protections of section 230, reinforcing the concept of an unmoderated public square. However, this opens up the question of algorithmic moderation. 

As summarized in a 2024 legal review by the American Journal of Comparative Law, “The Court thus avoided answering the question presented in Gonzalez--whether section 230 of the… [CDA] immunizes interactive computer services that make targeted recommendations” (72 Am. J. Comp. L. 247, 271 (2024)). This is a major point of contention in today’s legal communications, as little has been formally decided on the matter. In Force v. Facebook, algorithms are determined to at minimum be designed to be content neutral, and that they “take the information provided by Facebook users and “match” it to other users—again, materially unaltered—based on objective factors applicable to any content” (Force v. Facebook, Inc., 934 F.3d 53, 70 (2d Cir. 2019)). However, this assumed definition is beginning to be questioned as the concept of “shadowbanning” has come to the forefront of public discourse. As defined in Merriam-Webster, a shadowban is where a company “[causes] a user or their content to be hidden from some or all other users usually without the user's knowledge” (Merriam-Webster. (n.d.).). While difficult to prove, examples have been seen recently of this, used primarily against right-leaning accounts, but seen across the board. 

Algorithmic manipulation was recently defined and proven to fall outside of section 230’s legal immunity in Anderson v. TikTok Inc., a landmark case that very recently set a precedent and reestablished editorial control as a possible foundation for section 230 exemption. In this case, a young child perished while attempting a viral challenge seen and shared on the platform. Anderson sued and accused TikTok of continuing to recommend this and similar trends, in spite of clear knowledge of the harm they were both potentially and actively causing to their users. The court ruled that their direct control of the algorithm was “TikTok's own conduct, a subject outside of § 230(c)(1)” Anderson v. TikTok, Inc., 116 F.4th 180, 193 (3d Cir. 2024)). This decision holds potentially disastrous implications in the near future for Meta, Google, and other large corporations if current and future lawsuits are able to prove that these platforms are so much as directly manipulating their algorithm to promote specific viewpoints and suppressing others, let alone suspending or outright banning certain accounts without due cause. 

Section 230’s protection hinges on a platform’s neutrality in their content moderation, originally designed to essentially allow these platforms to remove patently offensive language and pornography. However, in addition to the “obscene, lewd, lascivious, filthy” (47 U.S.C. § 230(c)(2)) content-specific protection, the additional phrasing at the end of 230 has provided legal loopholes for companies to potentially exploit. “Or otherwise objectionable” is an outright subjective definition for content moderation and, it can be argued, is a significant oversight and flaw in the original language of this protection. Hypothetically, if both the “otherwise objectionable” and the "constitutionally protected” phrasing were removed, it would establish a clear and objective line between the designations of public-square and editorial publisher. In that case, a court would likely need answer only two questions to determine if a platform’s actions are protected: Did the platform take direct action, whether algorithmically or manually? And, if so, did the content in question fall under the definition of being obscene, lewd, etc?

If the answer to both of these questions is yes, then the platform would be protected under section 230. However, if it can be proven that direct action was taken in spite of the content not meeting the given definitions, then the platform would essentially have classified itself as a publisher, curating content based on their internal preferences. However, the addition of the final two specifications for offensive speech added significant subjectivity to the definition, allowing corporations to censor what they determined to be offensive, even if it would not meet the definition of “patently offensive”. This has led to a rash of First Amendment violation accusations, especially since the 2016 election. While few cases have been brought and decided on this matter, it has been demonstrated that platforms such as Youtube, Twitter (now X), Facebook, and others, established moderation guidelines that define “objectionable speech” to include misinformation – essentially, anything that the platform determines to be false based off of available evidence, or at minimum the common consensus. However, in the day and age of the internet, falsity has potentially expanded to include any spin, interpretation, or potentially unclear issue that might not match a corporation’s chosen values. 

As of the writing of this paper, a reasonable argument can be made that the ongoing court proceedings regarding InfoWars and Alex Jones are a recent major example of potential section 230 abuse. The given reasons for his suspension on almost every single social and content platform, simultaneously, were vague at best – Youtube, Twitter, Facebook, and others cited “hate speech and abusive behavior” as the primary reasons for his removal. Jones, a controversial conspiracy theorist, briefly suggested the possibility that the shooting was a manufactured tragedy (Lafferty v. Jones, 336 Conn. 332, 338, 246 A.3d 429, 437 (2020)). Despite his retraction of this statement and contextually tempering it as a rhetorical theory, it was deemed a violation of each platform’s community guidelines and he was effectively removed from the mainstream internet simultaneously (Jones, A. E., & Free Speech Systems, L. L. C. (2025, September 5). Petition for a writ of certiorari to the Supreme Court of Connecticut (No. 25-268)).  While his theories were unquestionably false and ill conceived, an act of mass coordinated suspension such as this sets a dangerous precedent for the internet going forward.

As of now, the interactive internet (social media, content sharing, etc.) has over five billion users, posting an estimated 500 million times per day on X/Twitter alone (Kepios, Digital 2026 report, 5.66 bn social media user identities; Wix, Social Media Statistics 2025)). A person’s reality is, no matter what, being more and more affected by how they perceive the world through the internet, so it is equally growing in importance that public-square platforms and relevant laws are studied with increased scrutiny. According to a study published in Proceedings of the National Academy of Sciences, a peer-reviewed scientific journal, it found that accounts and voices that didn’t line up with  a platform’s “desired average opinion” would be shadowbanned, censored, or outright suspended at a rate around three times higher than those that support it (Mosleh, Martel, Eckles, & Rand, 2024). Speaking hypothetically on a vertical spectrum, if a platform’s desired opinion landed halfway between the top and the middle, accounts that land anywhere between the middle and the bottom would be some of the most heavily silenced. Accounts that sit at the extreme top would also be affected, as both poles of the spectrum affect the overall average - in this case, however, bottom accounts far more than the top, as they make up a much larger percentage of users.

Section 230 has continued to exist as the consistent legal hurdle to anyone seeking to sue for First Amendment violations, and an unshakeable legal loophole for platforms seeking to avoid liability in the previously mentioned opinion reinforcement. A variety of cases were dismissed either partially or wholly on the grounds of 230 – Trump v. Twitter, Prager University v. Google LLC, Zuckermann v. Meta Platforms, and NetChoice v. Paxton. Each of these cases, while sometimes falling short due to lack of evidence, also were universally shut down on the basis of each platform’s first amendment rights of content moderation based on section 230. Additionally, the private and often closed-source nature of most mainstream public square platforms makes it supremely difficult to present enough empirical data to warrant a case being heard by a court, as proving actual account suppression becomes quite difficult without transparency in the moderation process (Trump v. Twitter, No. 1:21-cv-07040 (S.D.N.Y. 2021); Prager University v. Google LLC, 951 F.3d 991 (9th Cir. 2020); Zuckerman v. Meta Platforms, Inc., No. 1:24-cv-03697 (S.D.N.Y. 2024); and NetChoice, L.L.C. v. Paxton, 144 S. Ct. 1712 (2024)).

While the channels to do so are limited by the current web of legislation, a clear boundary for the range of platforms’ self-governance that is still protected by the First Amendment likely needs to be established. One of the few clear precedents in the media sphere for this is Miami Herald v. Tornillo, a 1974 case that established publisher’s rights and restrictions. The court’s decision ruled that a Florida law requiring newspapers to provide any critiqued figure equal space to respond was unconstitutional, leading to a full review of the rights of an editorial publisher (Miami Herald Pub. Co. v. Tornillo, 418 U.S. 241 (1974)). This case both provided a clear definition for a protected publisher, and was later applied in a much more recent setting, in the still-ongoing case of Moody v. Netchoice, LLC. This lawsuit challenged both Florida and Texas laws restricting content moderation, citing unconstitutionality. Each law essentially laid out transparency and action requirements in cases of moderation, requiring specific steps to be taken during the process of censoring or otherwise moderating a user’s content or profile (Fla. Stat. § 501.2041, 2021; Tex. Bus. & Com. Code §§ 120.001–120.104, 2021). The case was vacated and remanded back to the lower courts due to the fact that “neither the Eleventh Circuit nor the Fifth Circuit conducted a proper analysis of the facial First Amendment challenges to Florida and Texas laws regulating large internet platforms” (Moody v. NetChoice, LLC, 603 U.S. 707 (2024)).

So far, direct challenges to section 230 have been few and far between. The rapid evolution of technology is a difficult issue to address, as legislation and policymaking are far from agile. When section 230 – along with much of internet law – was written, few of today’s challenges involving technology and the absolute dominance of the web were expected or predicted in any way, shape, or form. It is somewhat of a surprise that 230 has remained untouched even today, having the rest of the legislation it was packaged in removed completely via court rulings. It remains to be seen what the outcome of Moody v. NetChoice might be; this case could potentially become a landmark case for mandated transparency within platforms claiming section 230 exemptions. If the ruling is in favor of the plaintiffs, it could potentially further expand protections for corporations, establishing that moderation requirements and public disclosure of methods and actions are unconstitutional. This would be a dangerous path to take and is likely the more controversial option, as enhancing the seemingly overbroad definitions within section 230. “[O]r otherwise objectionable, whether or not such material is constitutionally protected” (47 U.S.C. § 230(c)(2)) is a fascinatingly unique phrasing in light of modern legislation. Rarely, if ever, will one see constitutionally protected rights placed in the hands of private companies – which is, of course, the first step needed in possible reform. 

The public square as it existed decades ago no longer exists; the internet has fully replaced it in every way that matters. This establishes the main question surrounding free speech online: Where is the public square now? Very little mass communication happens beyond social media and content sharing platforms, as newspapers continue to slowly decline, mainstream news media loses its relevancy, and digital newsletters end up regularly sent to one’s spam folder. Simultaneously, the Meta/X/Google/TikTok set of platforms now make up a minimum of 85% of the internet’s public square by user reach and engagement time, with upper estimates reaching near 95% (Pew Research Center, 2024). Given this estimate, it is of great concern that there are extremely few public forums that are not owned by a company that, as of latest precedent, is free to moderate and censor as it chooses. So long as the subjective and broad phrasing of section 230 exists, free speech of any influence is subject to the whims of a biased algorithm, subtle or not. This is especially problematic with the recent advent of widespread artificial intelligence; social media algorithms are only as unbiased and objective as the programmers who created them, and the same goes – even more so – for generative AI. As of now, the only effective response from those seeking to challenge the status quo is to create alternative platforms with an opposing political bias to essentially balance out the internet. One of the more successful examples of this has been Rumble, a conservative counterpart to Youtube that seeks to limit moderation to extreme circumstances, along with the explosion of podcast-adjacent shows run by individuals, many of which still face algorithmic manipulation on one or more sites, but seek to cushion the effect by mass availability.

But creating new echo chambers is not the solution, as no platform will ever be free of bias – in spite of the purest intentions of the creators and moderators within. Section 230 needs to be amended, reformed, reinterpreted, or replaced completely. In the age it was created, “in good faith” was an olive branch of trust between legislators and private companies. And even if that trust was violated, the internet was far from having the monopoly on reality that it does now, so the consequences of a misstep by a platform were limited in scope and could be addressed as such. Good faith has been done away with, replaced by companies free to run with unfettered bias in their choice of narrative. Fortunately, we are early in this issue, and there are still several fairly uncomplicated routes to address it. Section 230 is often referred to as the “foundation of the internet”... and it is a foundation far too small for the skyscraper being erected upon it.

---

References

Anderson v. TikTok, Inc., 116 F.4th 180 (3d Cir. 2024).

Ashcroft v. Free Speech Coalition, 535 U.S. 234 (2002).

Brannon, V. C. (2024). Section 230: An overview (CRS Report No. R46751). Congressional Research Service. https://www.congress.gov/crs-product/R46751

Cubby, Inc. v. CompuServe, Inc., 776 F. Supp. 135 (S.D.N.Y. 1991).

Fla. Stat. § 501.2041 (2021).

Force v. Facebook, Inc., 934 F.3d 53 (2d Cir. 2019).

Gonzalez v. Google LLC, 2 F.4th 871 (9th Cir. 2021).

Jones v. Free Speech Sys, L.L.C., Petition for Writ of Certiorari, No. 25‑268 (U.S. Sept. 5, 2025).

Kepios / We Are Social. (2025, October). Digital 2026 Global Overview Report. We Are Social. https://datareportal.com/reports/digital-2025-global-overview-report

Lafferty v. Jones, 336 Conn. 332 (2020).

Merriam‑Webster. (n.d.). Shadow ban. In Merriam‑Webster.com dictionary. Retrieved November 13, 2025, from https://www.merriam-webster.com/dictionary/shadow%20ban

Miami Herald Publishing Co. v. Tornillo, 418 U.S. 241 (1974).

Moody v. NetChoice, L.L.C., 603 U.S. 707 (2024).

Mosleh, M., Martel, C., Eckles, D., & Rand, D. G. (2024). Shaping opinions in social networks with shadow banning. Proceedings of the National Academy of Sciences, 121(14), e2316078119. https://doi.org/10.1073/pnas.2316078119

National Telecommunications and Information Administration. (1995, July). Falling Through the Net: A Survey of the “Have Nots” in Rural and Urban America. https://www.ntia.gov/report/1995/falling-through-net-survey-have-nots-rural-and-urban-america-html

NetChoice, L.L.C. v. Paxton, 144 S. Ct. 1712 (2024).

Pew Research Center. (2024, August 14). Social Media Fact Sheet. https://www.pewresearch.org/internet/fact-sheet/social-media/

Prager University v. Google LLC, 951 F.3d 991 (9th Cir. 2020).

Reno v. ACLU, 521 U.S. 844 (1997).

S. Rep. No. 104-23, 104th Cong., 1st Sess. (1995). Telecommunications Competition and Deregulation Act of 1995. https://www.congress.gov/congressional-report/104th-congress/senate-report/23/1

Stratton Oakmont, Inc. v. Prodigy Servs. Co., 1995 WL 323710 (N.Y. Sup. Ct. May 24, 1995).

Telecommunications Act of 1996, Pub. L. No. 104-104, 110 Stat. 56 (1996).

Tex. Bus. & Com. Code §§ 120.001–120.104 (2021).

Trump v. Twitter, No. 1:21-cv-07040 (S.D.N.Y. 2021).

Twitter, Inc. v. Taamneh, 598 U.S. 471 (2023).

Wix. (2025). Social media statistics and facts. https://www.wix.com/blog/social-media-statistics-and-facts

Zuckerman v. Meta Platforms, Inc., No. 1:24-cv-03697 (S.D.N.Y. 2024).

47 U.S.C. § 230 (2024).

72 Am. J. Comp. L. 247 (2024).